Forecasting Spatio-Temporal Vegetation Changes in the Mealy Mountains Using a Cellular Automata-Markov Chain Hybrid Model
Zachery Bartlett (MSc) Geography
December, 10, 2011
Sub-arctic temperatures are expected to increase by approximately 4 deg. C by 2050. These changes are having impacts on vegetation patterns in arctic and sub-arctic environments, particularly along transition areas between forested and tundra ecosystems. Using multi-temporal satellite imagery, in combination with topographic variables, the changes in vegetation patterns from 1983 to 2008 were explored in a small, diverse region of the Mealy Mountains, Labrador. Bayesian probabilities were created for each land cover class, with topographic variables used as a priori additions to the probabilities. Vegetation changes were related to topographic variables, climate, and Bayesian probabilities. The Bayesian probability layers demonstrate the propensity for change of each land cover class used in the study. Knowledge of these changes was used in a cellular automata-Markov chain model to predict vegetation changes to 2020 and 2032. The predictions suggest movement of deciduous shrub along valley floors and into toe-slopes, as well as on protected, south-facing slopes. Coniferous shrub is expected to expand in the lower elevations (where it is dominant), and advance marginally along the valley floors.
The purpose of this study is to attain a better understanding of the effects of climate and topography on vegetation patterns, distributions, and movements. It expands on the body of research devoted to spatial-temporal modelling of vegetation. In a broader spectrum, the research complements the goals of the International Polar Year (IPY) research (2005).
The objectives of this study are to:
1. Create vegetation maps of historical land cover distributions;
2. Understand how land cover has changed given past climate fluctuations by comparing land cover distributions with known climate shifts;
3. Identify topographic conditions specific to the different land cover classes to aid in the development of a spatio-temporal forecasting model; and
4. Run a cellular automata-Markov chain model to forecast future land cover conditions as well as understand historic shifts in land cover.
Bounding Coordinates (N,E,S,W)
53.632, 53.563, -58.780, -58.895
Data not available until summer 2012
Permission of author
Dept. Of Geography, Science Bldg.
St. John's, NL